[GyuTaeBae]

Currently, at Compwin's(Plastic recycle company in Korea) recycle site, go through two rounds of crushing, weight screening, and cleaning processes to recycle the plastic. After these treatments, about 15 percent is recycled but, about 30 percent is disposed of again. This means that nearly half of the materials are not recycled and are thrown away again. In this Sludge, includes expensive materials such as flame retardant ABS (ABSF) and PCABS. Among them, the regenerative materials of the flame retardant ABS(used as heat-blocking material due to its low thermal conductivity) are high-value materials; 30% higher than the prices of ordinary ABS regenerative materials.

The problem is that it is hard to select flame retardant ABS from existing waste plastic processing processes. Because it's difficultly of locating ABS, some companies have expensive optics machine or expensive equipment to spot and to recycle ABS.(amounting to hundreds of millions of won - millions of dollars). But this method is not beneficial as it cost a lot to bring in those machines. What can I do to easily spot and detect the ABS materials? Can I use machine learning camera? or is there some other things that make ABS different from other plastics that I could use to spot them out?

So is there a sudden type of chemical marker that will only react with ABS? or is there another way to solve this problem?

[wildfyr]

I'll just say this, if anyone here knew that answer, they would be jogging... no... running to the nearest patent office to take the first step on making piles of money.

[wildfyr]

Really fast sampling Raman microscope? Very complicated to implement I'm sure.

[Enthalpy]

I'll just say this, if anyone here knew that answer, they would be jogging... no... running to the nearest patent office to take the first step on making piles of money.

  I know bizarre people who give ideas away for free 

[Enthalpy]

X-ray imaging machines exist that are sensitive to chemical elements, by using the K electrons' absorption edge. They serve to detect nitro-, nitroso- and nitrates in explosives. They would image nitrogen from ABS chips. Oxygen would distinguish from PA. The others have a density or composition very different.

Up to now, and from my very limited knowledge, such machines cost a bit, they have a limited geometrical resolution and reaction time, but your materials are very thin, which helps low-energy X-rays a lot.

The best person worldwide for that is Irène Dorion. If not retired, she may still be with
Smiths Heimann S.A.S.
36 rue Charles Heller
94405 Vitry sur Seine
France
Serge Maitrejean is few years younger.

My bet is that such a machine must be redesigned for your application. Your goal may be very easy for that technology, but the figures differ much from luggage screening.

Fun: I wanted to propose that application for years, but you asked before I described it.

Marc Schaefer, aka Enthalpy

[Enthalpy]

If the flame retardant contains bromine, phosphorus or antimony, then the X-ray machine easily distinguishes which ABS contains a flame retardant.

[Enthalpy]

Irène and co made imaging machines, but for plastic recycling, it seems easier to analyse one chip at a time and very quickly.

Some sort of V-shaped conveyor belt or whatever you want could put all chips in a line, let them fall at the end roll, where an air jet would separate the flame-retarded ABS from ABS, PE&PP, PET, PVC, others. A mono-pixel analyser before or after the roll would determine the opacity for C, N, O, P, Br, Sb and possibly more.

The speed limits I see are the brightness of the X-ray source and the time response of the air jet. I suppose some detectors are quick enough, but that's not obvious. The electronics and digital processing are faster than necessary. If needed, have several lines in parallel, as this looks cheap and easy.

Marc Schaefer, aka Enthalpy

[Enthalpy]

ABS is sensitive to solvents, more than PE&PP and PVC, and to different solvents than PA and PET.

I propose, not to dissolve the chips, but to soften the chips' surface with a solvent so they become tacky, which saves solvent and is much faster, especially if lukewarm. A solvent or a proper compound, like formic acid for POM. It could be a gas too, or a condensing vapour. A plasma, if selective enough? Glue manufacturers propose "primers" that are a first try.

Then, the tacky chips could adhere to some surface, for instance a conveyor belt, even when they're under the surface, or when air is blown on them, or they receive small jolts, or some delicate action.

It may help to throw the chips at the surface, for instance with an air jet, or a gaseous solvent jet.

The tackyfier could act at the surface or before. A wiper, jolts... would then harvest the tacky chips from the surface. Distillation could regenerate the tackyfier.

Varied tackyfiers separate the polymers specifically. Some other process should then separate the ABS chips that contain a flame retardant, for instance the suggested X-Ray process.

Marc Schaefer, aka Enthalpy

[Enthalpy]

As well heat can make the chips' surface selectively tacky. Supposedly, the polymer than softens first would become tacky. The Vicat point would be a first indication of that.

So the process could look like:

• Heat all chips, say in a moving gas or liquid or condensing vapour, to a first low temperature.
• Put all chips at a surface as previously suggested, possibly with impact, sort out the ones that adhere more.
• Repeat with a hotter temperature.

This one too seems a cheap method to separate all polymer families. It may fail with with very distinct sub-families, so a complementary separation before or after, say by flotation, would help.

==========

A variant would heat the polymers selectively by microwaves.

• PVC is very lossy and would heat first, so it would be separated as it gets tacky, as already suggested.
• PA is quite lossy too and would heat as the second separation candidate.
• ABS is more lossy than PE, PP and PET but not much. This would need a stronger microwave field, or a higher frequency, and the loss in ABS isn't quite reproducible. Removing ABS would be nice to pyrolysis too.
• Losses in PET and PE&PP are low. Possibly higher frequencies would heat PET selectively.

I don't expect this variant neither to discriminate flame retardants.

This method is very nice also for people willing to pyrolyze plastics to fuels, since little PVC in PP&PE makes the process much more polluting. Removing ABS too is desireable. This could be done at once, leaving only PE&PP and possibly PET.

Marc Schaefer, aka Enthalpy

[Enthalpy]

As well surface hardness might let sort plastics.

For instance with an air blow, throw the plastic chips at an item covered with needles. With adjusted conditions, the softer chips should stick at the needles, the harder not.

The chips could be thrown directly under the selective surface which moves to a wiper to harvest the stuck chips. The surface can be a portion of a disk.

Optionally, heat by contact or by microwaves, or the contact with some fluid, possibly warm, can soften the chips' surface selectively.

If the selectivity doesn't suffice, as the chips' orientation can mess up, several steps like in distillation towers or centrifuges improve it.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Permittivity and its losses lets distinguish polymer chips to sort by other means, for instance an air jet with fast reaction.

Take 10mm×10mm×10mm between parallel plates: the roughly 85fF resonate at 2.45GHz with two single turns of D=60mm which, of gold-plated D=10mm, bring Q≈10 000. A plastic chip of e=2 and 2% losses that passes through and occupies 10% of the volume detunes the resonant circuit by <5%, giving a fist idea of the chip volume, and above all it dampens the resonance to Q<1000, easy to measure, if needed within <20µs.

The machine would throw successive plastic chips through a pipe, for instance of alumina, at a rate limited only by the reaction time of the sorting actuator, which could have a rotating barrel to give many valves more time to react.

Here too, PVC could be sorted out first, followed by PA, and then ABS would show the biggest losses, while PET can be distinguished from PE&PP. The proper actuator(s) would sort all polymer families after a single measurement.

Higher frequencies are possible, they distinguish PET better and allow closed metal sections as wave guides and resonators, but most low-loss plating metals are sensitive to erosion. Palladium maybe.

At 1ms rate and 5mm×5mm×5mm chips, the machine would sort 0.12L/s or 0.45m³/h, so faster actuators are desired. Or build 1000 channels in parallel and sort 450m³/h = 7.5m³/min = 0.12m³/s, satisfying.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Two messages ago, I proposed to throw the plastic chips on needles.

A possibly better alternative would press the chips against the needles with a limited force. The chips would rest at, possibly on, an elastic surface that nears the surface covered with needles, then goes away.

An endurant polymer like PU could make the elastic surface. Steel is excellent but it looks less easy. Making the needle-bearing surface elastic, or the needles themselves, seems more difficult.

Possible shapes include two rotating converging cone frustrums, the lower one elastic that brings all chips to the interaction zone, the upper one covered with needles to pick the softer polymers, with a wiper or a rotating brush to harvest the selected chips.

Marc Schaefer, aka Enthalpy

[Enthalpy]

Scientists should never use figures. They spoil the very best ideas.

The appended diagram made by the spreadsheet (change its .pdf to .xls) tells the opacity of the target materials versus X-ray photon energy. Multiply the cm²/g of an element by how much g/cm³ the material contains, and you get the contribution to the exponential attenuation factor. Many thanks to Nist
http://physics.nist.gov/PhysRefData/XrayMassCoef/tab3.html
http://physics.nist.gov/PhysRefData/XrayMassCoef/tab4.html

The navy peaks locate the (set of) emission line added by an element to the continuous background when it serves as an anode. This transition to the K level, or also L on the diagram, is the reciprocal process to the stronger absorption of photons exceeding an energy threshold, except that the electron source is more diverse. X-ray tubes can have an exit filter of an element but heavier than the anode to attenuate energies beyond the threshold and dampen the less penetrating lower energies. I've put the refractory (semi-) metals, including those expensive or hard to separate, and as they operate in vacuum, even the easily corroded ones. The exit filter at room temperature accepts compounds of varied elements.

Flame-retardant bromine is easily detected in the target polymer chips.

• W emits at 12.1keV by its transition to L1. Re, Os or Ir can filter it.
• Br absorbs stronger above 13.5keV by its K electrons.
• Pt emits at 13.7keV, Au can filter. Or Au/Y for 14.2keV.
• Polyethylene attenuates 12.1keV as exp(-x/8.6mm), nearly perfect for chips supposedly 2-3mm thick. Air is nearly transparent.
• 1wt% Br attenuates 1.5× more than PE at 13.7keV but 0.2× more at 12.1keV. Other elements follow the usual slope. Compare both attenuations, done.

The irregular chips must keep their orientation between the measurements. A conveyor belt, not a free fall. The sources and detectors calibrate when no chip is present.

Antimony is less easily detected.

• At 24.1keV from Pd/Ag, 1% Sb adds 0.39 and 1% Cl 0.12 to PE's attenuation.
• At 42.0keV from Pr/Gd, 1% Sb adds 0.90 and 1% Cl 0.20.
• At 65.4keV from Hf/Ta, 1% Sb adds 0.28 and 1% Cl 0.02. This system is not too badly conditioned.
• 3mm PE attenuate 65.4keV ×0.95, so the signal is in the ×0.05 variation, uncomfortable.

An expert would possibly prefer to use the continuous spectrum of W filtered by Sn and by Te. Despite being more fuzzy, the spectra may give information more specific to Sb.

PVC chips are easy to recognize. At 8.3keV from Ni/Cu, they are 12× as opaque as PETP, accepting some tolerance on the thickness. A vertical line of sight, with a thin belt, eases that. Or check the attenuation ratio between 12.1keV and 65.4keV.

Flame-retardant chlorine presents no usable threshold. Its K absorption starts at 2.8keV, where 3mm PE attenuates by 10¹⁰. The knee between 20keV and 50keV seems most informative, as described for antimony.

Did I maybe perhaps write "detect nitrogen and oxygen from PA and ABS"? No usable K threshold allows that. Some information is hidden in the 20keV to 50keV knee if measuring at several energies there. This system isn't well conditioned, so the chip must not move, and the detectors be well aligned. Not simple.

==========

X-ray fluorescence doesn't need the photons to traverse the full chip, so it can use more varied energies. Flame-retardant chlorine is easily detected. An expert would probably have have chosen that right from the beginning.

Marc Schaefer, aka Enthalpy

[wildfyr]

I like the X-ray approach to catching bromine FR materials.

I should also note though, that phosphorous based materials are common, and increasingly displacing halides as FR materials due to environmental considerations.

[Enthalpy]

Thanks Wildfyr!

Phosphorus would be about as uncomfortable to detect as chlorine: from attenuation at several energies around the 20-50keV knee, or by fluorescence.

Distinguishing P from Cl would be one step more complicated, but maybe the recycling application doesn't need it. GyuTaeBae only spoke from separating the chips that contain some flame retardant.